Today at the IGF, I sat on the panel for discussions on IPv4 to IPv6 transition. The statement I wanted to make was below, but it was apparently a bit long as I got cut.
Thank you Madam Chair, Ladies and gentlemen, good morning
I'd like to share with you some of the issues, and developments from an operational and operator perspective with regard to the transition from Internet Protocol version 4, IPv4 to Internet Protocol version 6, IPv6. These protocols are not interoperable, but can be run in parallel. For an every day user, this means that if you only have access to IPv4 or only IPv6 and try to reach content available only in the other protocol, this will not work. But more on the issues of transition later. First of all, why would we need this, and what are the underlying issues. Simply put, the original design of the Internet, was not meant to scale to the size of today. Luckily however, the Internet architecture allows for each layer in the architecture to change or be replaced - but perhaps not as easily as originally thought.
But let's first look at what the obstacles are for migration, from a provider point of view. The most often cited obstacle is that there is no customer demand, customers want to run peer-to-peer applications and access web-sites. They have no notion of IP transport and don't really care - and shouldn't have to. IPv6 does not enable new services that can't be delivered today, but it does address the upcoming shortage of IPv4 addresses.
So what are operators doing? Deployment of IPv6 capable equipment is, and should be incremental as operators are procuring and deploying new hardware and software in their normal upgrade cycles. The reason for this, obviously, is lack of demand from customers that means there is no added revenue, therefor requirements for forced upgrades and/or change in order to deploy IPv6 will make no economical sense, at least in the short term.
Further, operators in developed as well as the developing world more or less face the same challenges, lack of commercial drivers and faced with upgrades. The same applies to the end-users, enterprises and residential equipment. In the developed world we face the upgrade of software and in worst case hardware for computers and further all cable-TV modems, DSL-modems, home gateways, firewalls etc. Here it might even be that the developing world will be at an advantage.
Current core equipment however, i.e equipment used in the backbone, that operators are procuring and to some extent already using, is supporting IPv6. Issues that remain and is becoming urgent is operational experience, and "debugging" of software and configurations. Also, building field experience will show what operational management tools are needed to be developed, or migrated to IPv6 support. In our deployment we found that while products supported IPv6 as in forwarding of packets, they often lack in parity when it comes to operationally needed tools. Vendor pricing, in the form of extra charges for IPv6 support, has also not helped in the deployment. But here there are positive signs.
In order to cater for a scenario where operators have migrated their core networks, but where endusers might still be on legacy equipment running IPv4, or for customers on an IPv6 network wanting to access content that is still only available on IPv4, the Internet Engineering Task Force, IETF - the Internet standards organization, is working on technology for bridging the two protocols.
Currently, the most pressing need is to get products for customer premises equipment, that supports IPv6. DSLmodems, cable-modems, middleboxes etc are the most urgent category that needs to get IPv6 support, and so far IPv6 support in this area has been very poor. The good news is that it seems vendors are starting to realize they are the last component to provide IPv6 support. With this, and with the back-office system vendors starting to complete migrations to IPv6 support, the operator networks are migrated.
Most day-to-day modern software, supports IPv6. Enterprise applications might still need to be migrated, but even here there are positive signs as software vendors are upgrading their products. Legacy applications will however be around for the foreseeable future, which is another driver for providers and the IETF to work on technology to support them in parallel and in bridging them to an IPv6 enabled world.
The bridging features between IP4 and IPv6 will be a crucial component in a fast, and smooth deployment to IPv6, both for providers and end-users. And hopefully, will also take away some of the contention around the run out of IPv4 addresses.
So, in these economically uncertain times, will operators be willing to bear the cost of the migration? As the investments needed are mostly taken as part of the normal upgrades cycles, that will be less of an issue. Training and operational maintenance will indeed be costs, but on going training is part of operators planning in any case. Where costs might have to be incurred is on the software side, where systems might have to be upgraded or replaced. But again, I would argue that it is hard to separate true cost of migration from normal upgrade cycles. And much has been upgraded as part of the on going upgrade cycles. For the end users and enterprises, the migration and investments will be slower, which is again why a translation technology will be needed, and is being worked on.
Last, so is there any active mass deployments? Many are deploying in their core networks, and working on customer facing systems as well as back-office systems. But one real example is FreeTelecom in France, that by the fact they are provisioning their own own CPE equipment, could enable IPv6 for all of their 3 million subscribers. In a first step limited deployment, it's available free of charge to their triple-play customers, of which 250,000 customers have actively asked to enabled it. Which in relation to the total of 3 million users, might seem like a small number, but on a global scale it is a stunning number. So there is indeed deployment starting to happen.
Thank you!